**4:**On line 9, the index of θ and p should be k, not n. [Thanks to Amir Alipour]**4:**In Example 1.3, in the last line, the inequality n>q should not be strict. [Thanks to Amir Alipour]**5:**In Formula (1.5), the power of σ should be n+q+1. [Thanks to Amir Alipour]**7:**In Formula (1.7), the parameters a and b should be exchanged. [Thanks to Amir Alipour]**8:**After Formula (1.10), the gradient symbol ∇ should be defined, instead of waiting for page 19. [Thanks to Amir Alipour]**8:**In Example 1.6, the cumulant function ψ should involve log(-1/2θ_{2}). [Thanks to Amir Alipour]**10:**In the definition of the Bessel function at the top, the series is in t not in z. [Thanks to Amir Alipour]**10:**In the last formula of Example 1.9, the likelihood function is inversely proportional to a power of the displayed product and the power of σ should be 2n/(p+1). [Thanks to Amir Alipour]**16:**In the Challenger example, the correct reference is to Example 7.11, not Example 7.3 [Thanks to Patrick Hung]**38:**In the transform Y_{n}of X_{n}, the argument of the arcsin should be 2X_{n}-1 instead of X_{n}[Thanks to Patrick Hung] and F is not defined as the arcsine transform in Fig. 2.1 [Thanks to Gholamhossein Gholami]**49:**In the denominator of the integral ratio at the bottom of the page, the outer integral should be on X, not on A [Thanks to Daniel Bové]**52:**In Example 2.19, we are looking at the minimum of M in b, not the maximum, and the power of (1-b) is -(α-a) [Thanks to Gholamhossein Gholami]**53:**In Example 2.20, the ratio f/g_{α}should start with exp(α(z-μ)) and the optimum (2.11) in α is missing a 1/2 in the first μ [Thanks to Gholamhossein Gholami]**54:**In Example 2.22, there is a (1/ √ 2 π) missing in the inequality and the bound is useless when x^{2}is greater than 2, not lower, which occurs with probability 0.39 [Thanks to Daniel Bové]**55:**In Algorithm A.6, α should be equal to λ and k to log(β) - λ - log(c) [Thanks to Daniel Bové]**61:**On Figure 2.8, the labels e^{a1+b1x}and e^{a3+b3x}have to interchanged [Thanks to Daniel Bové]**62:**The simulation from g_{n}is incorrect: we need to simulate a truncated exponential on one of both subintervals of [x_{1},x_{2}] [Thanks to Daniel Bové]**63:**In Problem 2.8, question (b), the ratio in arctan should be`X`_{2}`/X`, rather than_{1}`X`/X_{1}_{2}[Thanks to Wolfgang Remmel]**64:**In Problem 2.11, question (b), the Cauchy distribution function is tan^{-1}(x)/π + 1/2 [Thanks to Mouhamad Mounirou Allaya]**80:**In Example 3.1, the reference to Example 1.7 should be to Example 1.8 and λ should be defined as ||θ|| to correspond to the polar coordinates [Thanks to Daniel Bové]**86:**In formula (3.5), a factor 2 is missing at the beginning [Thanks to Daniel Bové]**92:**In the last formula, √(2πσ) should be √(2π)σ [Thanks to Daniel Bové]**94:**In the definition of f_{Y}, (x-4.5) should be (y-4.5) [Thanks to Daniel Bové]**98:**In the definition of g, the ≥0 sign should be a subscript [Thanks to Daniel Bové]**106:**In the bound M', α and a should be exchanged [Thanks to Daniel Bové]**114:**In Problem 3.22, the sum of the weights ω_{i}should be equal to n in expectation, not to 1 [Thanks to Daniel Bové]**126:**In formula (4.5), all the powers should be outside the expectations [Thanks to Vadim Victorovich Zipunnikov]**127:**In the footnote, open should opened**128:**On line -8, the X_{i}'s should be normal with mean µ not mean 0*[Thanks to Matt Wolak]***129:**The last diagonal term of the matrix (4.7) is n^{2}not^{2}*[Thanks to Tomas Ukkonen]***178:**Both terms in the first inequality are the same! The numerator on the rhs should be k(Z|θ_{(j)},x) .**178-179:**The details of Example 5.17 are full of typos: there is a minus sign missing in the first and second equations in front of (z_{i}-θ)^{2}, the indices should start at m+1 in the vector z, the density (5.16) is incorrect, missing the normalising factor (1-φ(a-θ))^{m-n}, the indices should start at m+1, not (n-m+1) in the second sum of the last equation of page 178, and, in the update equation (5.17), the (1/n) in the last term should be (n-m)/n... [Thanks to Daniel Bové & Serena Ng]**192:**In connection with the above, Problem 5.16 (b) is missing a +θ' in the expectation of Z_{1}[Thanks to Daniel Bové & Serena Ng]**221:**In the proof of Proposition 6.31, the atom is Cx{1}, not Cx{0}*[Thanks to Vivekananda Roy]***282:**The prior in the third line puts an exponential prior on exp(α) not on log(α) [Thanks to Uli Schneider]**298:**In the construction of w^{j}_{i}, the index on`ρ`should be ji_{tj}**302:**In Problem 7.1(a), the Gamma(4,7) should be a Gamma(4,6) [Thanks to Yu Huang]**308:**In Problem 7.25 (a), it is exp(α) that has an exponential distribution, not log(α) [Thanks to Uli Schneider]**319:**In the first equation, the ratio of exponentials should be inverted [Merci, Olivier Cappé]**365:**In Algorithm A.38, the scale parameter in the simulation of t^{2}_{j}is wrong: it is missing a +n_{j}x^{2}_{j }[Thanks to Justin Veenstra]**385:**The conditional distributions on the`C`'s are not normal, due to the normal prior on the_{i}`log C`'s_{i}**429:**In the second paragraph, there is a missing "corresponding to" between posterior distribution and (11.2) [Thanks to Daniel Bové]**431:**In the last Jacobian, θ_{n}^{(t)}should be θ_{m}^{(t)}[Thanks to Björn Bornkamp]**432:**In the Jacobians, the determinant is missing and at the bottom of the page "it can solved" should be "it can be solved" [Thanks to Daniel Bové]**433:**In the Jacobians, there is one parenthesis too many in the denominators [Thanks to Daniel Bové]**436:**In the Jacobian, p^{(3)}_{2}should have a minus sign but this does not change the final value [Thanks to Quan Wen]**438:**In the definition of σ_{(j+1)(k+1)}on the rhs, the square is missing [Thanks to Mauro Bernardi]**439:**In the second formula, there is a missing bracket after θ_{k+1}[Thanks to Mauro Bernardi]**476:**In Example 12.8, in the numerator of w_{x}, p^{1/2}should be replaced by p a^{1/2}*[Thanks to Vivekananda Roy]***497:**In the definition of both`B`and_{T}`W`, it is more standard to use unbiased estimators, namely to use M-1 rather than M and T-1 rather than T, respectively [Thanks to Uli Schneider]_{T}**528:**In Algorithm A.56, there is an approximation in the recycling of w^{(-t)}_{i}that creates a bias, as explained in this technical report [Thanks to Josef Leydold & Wolfgang Hoermann]**564:**In Algorithm A.62, part c., the weights in the denominator should be`ζ`not_{k}`ρ`[Thanks to Iain Murray]_{k }**575:**In Problem 14.16 (b), there is a missing plus sign between both terms in the rhs [Thanks to Mauro Bernardi]**583:**In the Negative Binomial Distribution in A.13, n+x+1 should be n+x-1 [Thanks to Gholamhossein Gholami]

Last modified on Oktober 16, 2009.